What the Nexcom Local News Index measures, how it is measured, and the specific ways it can mislead you. Read this before citing it.
A market is a city where Nexcom Media operates a newsroom. That is the selection rule, and it is not random: markets are here because Nexcom chose to publish in them, not because they were sampled to represent the country. 40 markets across 8 states and 27 counties are included in the 2026-07 edition.
An outlet counts when Nexcom monitors a machine-readable feed (RSS/Atom) it publishes about that market. Each is hand-verified at the time it is added and typed as television, newspaper/digital, government, or weather service. Aggregator feeds (for example Google News queries) are excluded from outlet counts, because they are not newsrooms.
The bias this creates: a newsroom that publishes no feed is invisible to this index. That skews against small independents, newsletter-first outlets, radio, and anything paywalled without a feed. Outlet counts here are a floor, not a total. Do not read a low number as "this city has few newsrooms" — read it as "few feeds were reachable."
Articles are those Nexcom has fetched and stored from tracked feeds: a live rolling window plus an archive. The 2026-07 edition covers 151,864 articles. Syndicated wire stories carried by several outlets are counted once per outlet that published them, so shared regional coverage inflates totals relative to original local reporting.
Every feed in the register is re-tested each edition. On 2026-07-18 we fetched all 237 endpoints with a self-identifying agent and recorded, per feed, the HTTP status, whether any item parsed, and the date of the newest item. Each feed is then labelled:
| Status | Count | Meaning |
|---|---|---|
| Publishing | 201 | Returned a parseable item dated within 30 days |
| Blocks our fetcher | 12 | Returned HTTP 403 or 429. This almost always means the outlet blocks automated clients, not that it has stopped publishing. We never record these as closed. |
| Unreachable / undated / stale | 24 | URL did not resolve, or had no machine-readable dates, or newest item was over 30 days old |
84.8% of tracked feeds were publishing. A feed being unreachable to us is not evidence an outlet has closed — only that its feed was not machine-readable from our vantage point on the test date.
Feed accessibility splits sharply by outlet type:
| Outlet type | Feeds | Blocked our fetcher |
|---|---|---|
| Television | 120 | 0 (0%) |
| Newspaper / digital | 90 | 12 (13.3%) |
| Government | 24 | 0 (0%) |
Every blocked feed in this edition belongs to a chain-owned newspaper. What this does and does not show: it shows that the newspaper groups in these markets configure their feed servers to refuse automated clients, while the television stations do not. It does not show that newspapers publish less, or that their journalism is worse — only that their feeds are less machine-accessible. Cite it as a statement about feed accessibility, not about news output.
32 of 132 unique feed hosts appear in more than one market. In the Houston suburbs, for example, the same metro television feeds (abc13.com, khou.com) serve eight markets each. This means "outlets tracked" in a suburban market is partly metro coverage that also reaches neighbouring suburbs, not eight independent newsrooms. The per-market tables mark how many of a market's feeds are shared with other markets.
Topic labels come from a deterministic keyword classifier, not human coding and not a language model. Each article's headline and summary are tested against keyword sets in a fixed order — safety, government, schools, parks, library, business — and the first set that matches wins. Anything matching nothing is labelled local.
We audited the classifier against 700 live articles drawn from 14 markets. The results:
| Outcome | Share | What it means |
|---|---|---|
| Matched exactly one topic | 36.6% | Label is unambiguous |
| Matched several topics | 7.9% | Priority order chose the label; 59 competing labels were suppressed |
| Matched nothing | 55.6% | Defaulted to local |
Three consequences you should carry into any citation:
local share mostly measures what the taxonomy fails to describe. The taxonomy meaningfully classifies roughly 44.4% of content.Topic percentages are therefore a coarse directional signal about emphasis. They are not a content analysis, and they should never be quoted as "X% of local news in this city is about Y." The defensible phrasing is "X% of articles Nexcom indexed in this market were classified as Y."
Nexcom Media operates a newsroom in every market this index measures, and the tracked outlets include organisations Nexcom competes with for audience. We publish the index anyway because the underlying counts are verifiable — the outlet lists and article totals are reproducible from the public API — but you should weigh the source. This index is not an independent audit of competitors, and nothing in it should be read as an assessment of any outlet's quality.
Not all figures here are equally solid. In descending order of confidence:
Each edition is stamped with a year-month (2026-07) and a generation timestamp (2026-07-18T21:58:50Z). Editions are snapshots: article counts move continuously, so a citation should name the edition. Prior editions remain reachable at their dated download URLs.
If an outlet is missing, mis-typed, or wrongly attributed to a market, write to editor@nexcom.media and we will correct it in the next edition and note the change. See also our corrections policy.